IT Management Weekly Overview — Week of November 16–November 21, 2026

IT resilience, vendor risk, and accountable AI

An editorial overview of the week’s key themes in IT Management


If one word defined this week in IT management, it was dependency. Whether the subject was a chip supplier in turmoil, a cloud outage, or a sprawling estate of AI agents, the posts kept returning to the same uncomfortable question for CIOs: who actually controls the systems the business now runs on, and what happens when that control turns out to be thinner than the org chart suggests?

The Nvidia shock and the limits of concentration

The biggest story of the week was the leadership upheaval at the world’s most important AI supplier. Jensen Huang’s exit from Nvidia exposes a paradox at the heart of sovereign AI strategies: nations pursuing technological independence have, in many cases, simply swapped one dependency for another, building national infrastructure on a single vendor’s GPUs, networking and CUDA ecosystem. Governments may now push harder on architectural diversity and domestic chipmaking.

For enterprises, the lesson is closer to home. What the Nvidia crisis reveals about enterprise vendor risk is that governance failure at a critical supplier quickly becomes a continuity, compliance and concentration-risk problem for its customers. The real test is whether existing access rights, contracts and exit plans would survive a supplier stumbling. Compounding the pressure, reporting that Nvidia privately confirmed it could track chips to fight smuggling suggests GPU compliance controls are likely to tighten, with location verification and audit scrutiny landing on organizations whose asset visibility may not be ready.

The same pattern surfaced outside the AI hardware stack. The Microsoft outage demands an end to vendor concentration risk, arguing that consolidating identity, collaboration and security on one provider is a management decision, not just an availability issue. Leaders need to decide which capabilities deserve funded fallbacks and tested recovery playbooks before commercial convenience hardens into operational fragility.

AI governance: from titles to teeth

The second major thread was a sustained critique of how organizations govern AI. Several posts converged on the idea that policies and job titles are not the same thing as control. AI governance’s authority gap explains why appointing AI leaders hasn’t solved accountability: authority, accountability and technical enforcement are often disconnected, and insurers, vendors or incidents will expose that gap if CIOs don’t test it first.

In a similar vein, the argument that AI governance has a strategy problem points out that many firms have AI policies that never touch funding, approvals or operating decisions. The remedy is tiered governance with clear decision rights, especially for agentic AI, scaled to business risk.

The most practical framing came from the case for building trustworthy AI by fencing it in rather than predicting it. Instead of trying to forecast what a model will do, IT leaders should decide where AI may recommend, decide or act, and enforce safeguards outside the agent’s reach. It is a refreshingly concrete answer to an abstract debate.

Scaling without sprawling

Governance gaps become most visible as AI moves from pilots to production. AI has a scaling problem recasts scale as a portfolio and operating-model challenge: every funded initiative should strengthen reusable platforms and measurable value rather than add another isolated use case. That theme carried into building AI that can work across the enterprise, which stresses standardizing data, governance and funding discipline before agents enter business-critical workflows.

The cost of ignoring this advice already has a name. Stop the sprawl: 5 steps to wrangling AI orphans treats agent proliferation as an identity and lifecycle problem, warning of orphaned agents with excessive permissions, duplicated spend and nobody accountable when something goes wrong.

Risk ownership beyond the data center

Finally, two posts widened the lens on who owns risk. Asking who should be on a CIO’s IT risk management team surfaces a sobering idea: the biggest exposure may be fragmented accountability rather than any single vulnerability, and people-focused controls deserve the same funding as technical ones. Meanwhile, the call to define your business’s operational technology before attackers do urges leaders to identify which connected systems truly run the business and govern OT security by business impact, not outdated asset categories.


Taken together, the week’s reading delivers a consistent message. Concentration, whether in a chip vendor, a cloud platform or an unmanaged fleet of AI agents, is a choice that accumulates quietly until a crisis makes it visible. The CIOs best placed for 2027 will be those who map their dependencies honestly, connect governance to real decision rights and enforcement, and fund resilience before they need it.


Full post index for this week:

Browse the full IT Management archive at genesis-aka.net/information-technology/management/

Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.

Leave a Reply